Add a Settings.FilterOnLoadTime histogram to observe time spent on this critical path in the wild.

Add a Settings.FilterOnLoadTime histogram to observe time spent on this critical path in the wild.

This call is on the startup path; we believe all regressions have been addressed, but would like data from the wild to see how this truly behaves in various user configurations.

BUG=330908
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=247893

[gab, 2014-01-28 23:26:18]

Nico/Alexei, PTAL.

Thanks!
Gab

[Alexei Svitkine (slow), 2014-01-29 14:33:40]

lgtm

[gab, 2014-01-29 14:57:37]

On 2014/01/29 14:33:40, Alexei Svitkine wrote:
> lgtm

Thanks @Nico for OWNER.

Alexei: UMA_HISTOGRAM_TIMES is the shortest one I could find (1ms to 10s), but I
don't expect this to ever go much above 100ms; will this still provide precise
data in the 1-100ms buckets (while still being able to report big spikes should
they occur)?

[Alexei Svitkine (slow), 2014-01-29 15:33:30]

On 2014/01/29 14:57:37, gab wrote:
> On 2014/01/29 14:33:40, Alexei Svitkine wrote:
> > lgtm
> 
> Thanks @Nico for OWNER.
> 
> Alexei: UMA_HISTOGRAM_TIMES is the shortest one I could find (1ms to 10s), but
I
> don't expect this to ever go much above 100ms; will this still provide precise
> data in the 1-100ms buckets (while still being able to report big spikes
should
> they occur)?

Yes, that macro should be fine for this use case. Histogram macros (other than
enum and percent ones) produce a bucket distribution where the buckets are
exponentially increasing in range.

The easiest way to see the distribution is to find an existing histogram that
uses that macro on the dashboard. For example, in this case:

https://uma.googleplex.com/histograms/?dayCount=1&endDate=01-28-2014&version=...

[Alexei Svitkine (slow), 2014-01-29 15:37:33]

On 2014/01/29 15:33:30, Alexei Svitkine wrote:
> On 2014/01/29 14:57:37, gab wrote:
> > On 2014/01/29 14:33:40, Alexei Svitkine wrote:
> > > lgtm
> > 
> > Thanks @Nico for OWNER.
> > 
> > Alexei: UMA_HISTOGRAM_TIMES is the shortest one I could find (1ms to 10s),
but
> I
> > don't expect this to ever go much above 100ms; will this still provide
precise
> > data in the 1-100ms buckets (while still being able to report big spikes
> should
> > they occur)?
> 
> Yes, that macro should be fine for this use case. Histogram macros (other than
> enum and percent ones) produce a bucket distribution where the buckets are
> exponentially increasing in range.
> 
> The easiest way to see the distribution is to find an existing histogram that
> uses that macro on the dashboard. For example, in this case:
> 
>
https://uma.googleplex.com/histograms/?dayCount=1&endDate=01-28-2014&version=...

Sorry, here's a better link:

https://uma.googleplex.com/histograms/?dayCount=1&endDate=01-28-2014&version=...

(Limits to a specific version, so less noise in bucket ranges due to
corruption.)

[gab, 2014-01-29 16:40:59]

Sounds good thanks, also added an identical histogram for
Settings.FilterSerializeDataTime.

PTAL.

Thanks!
Gab

[Alexei Svitkine (slow), 2014-01-29 16:42:27]

lgtm

[Nico, 2014-01-29 17:58:12]

lgtm

https://codereview.chromium.org/149073003/diff/40001/base/prefs/json_pref_sto...
File base/prefs/json_pref_store.cc (right):

https://codereview.chromium.org/149073003/diff/40001/base/prefs/json_pref_sto...
base/prefs/json_pref_store.cc:322: // startup.
Put a date on this ("remove by Feb 14" or similar) and below.

[gab, 2014-01-29 19:15:55]

Thanks, done.

https://codereview.chromium.org/149073003/diff/40001/base/prefs/json_pref_sto...
File base/prefs/json_pref_store.cc (right):

https://codereview.chromium.org/149073003/diff/40001/base/prefs/json_pref_sto...
base/prefs/json_pref_store.cc:322: // startup.
On 2014/01/29 17:58:13, Nico wrote:
> Put a date on this ("remove by Feb 14" or similar) and below.

Done, used Feb 21, might be before if the data appears conclusive.

[commit-bot: I haz the power, 2014-01-29 19:18:00]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/gab@chromium.org/149073003/60001

[commit-bot: I haz the power, 2014-01-30 08:16:28]

Change committed as 247893